Comparison of Costello Geomagnetic Activity Index Model and JHU/APL - - PowerPoint PPT Presentation

Comparison of Costello Geomagnetic Activity Index Model and JHU/APL Models for Kp Prediction

By: David Marchese Mentors: Douglas Biesecker Christopher Balch

Outline

Background Kp Prediction Costello Geomagnetic Activity Index Model

Validation Studies

Research Results JHU/APL Models Conclusions

Kp Index

• Developed by Julius Bartels
• Measure of the maximum disturbances in the

horizontal components of Earth’s magnetic field caused by solar particle radiation

• Official index calculated every three hours

using observations from 13 subauroral magnetometer stations

Kp Values

Range from 0 to 9 in a

scale of thirds

Kp value of 0

corresponds to the quietest conditions

Kp value of 9

corresponds to the most disturbed conditions

Quasi-logarithmic

scale

ap index ranges from 0

to 400 and represents the Kp value converted to a linear scale in nT

Effects of Geomagnetic Storms

• Disrupt radio communications
• Disrupt GPS navigation
• Damage transformers and electric power grids
• Degrade satellite instrumentation
• Increase satellite drag
• Cause aurora
• Confuse racing pigeons

NOAA Space Weather Scales

• NOAA G-Scale based on Kp estimates

from the Boulder-NOAA Magnetometer

• Warnings issued when Kp values of 4, 5,

6, and 7 or greater are expected

• Alerts issued for Kp values of 4, 5, 6, 7, 8,

and 9

NOAA G-Scale

USAF Estimated Kp

Official Kp index published

with significant time delay

“Nowcast” Kp algorithm

provides real-time estimates

• f Kp

Derived using data from 9

ground-based magnetometers in North America

Calculated by the United

States Air Force 55th Space Weather Squadron

Costello Geomagnetic Activity Index

• Neural network algorithm trained on the

response of Kp to solar wind data

• Input two hours of data for solar wind

speed, IMF magnitude, and Bz

• Output running 3-hour Kp every 15

minutes

Motivation for Research

Space weather forecasters need to know how reliable

prediction models are

Several validation studies have been done on the

Costello model

Results are not complimentary Important to determine the reasons for discrepancies

Costello Validation Study 1

• Covers the time period from August 17,

1978 to February 16, 1980 (ISEE-3)

• Predictions binned to integer values

between 0 and 7

• Tends to underpredict high and low Kp

values

underprediction

• verprediction

Study performed by members of the Space Environment Center.

Costello Validation Study 2

• Covers the time period from 1975-2001

(IMP-8, Wind, ACE)

• Official Kp values obtained by

interpolating between points to match 15 minute time granularity

• Tends to overpredict low Kp values and

underpredict high Kp values

• Correlation coefficient = 0.75
• verprediction

underprediction

Study performed by Wing et al.

Research

• Find the distribution of official Kp values for a given prediction
• Determine if the models perform differently during solar maximum years than

during solar minimum years

• Compare the performance of the Costello model to the JHU/APL models
• Data Set
• Supplied Costello prediction data spans

from July 1, 1998 until June 18, 2007

• Data gap from May 7, 2005 until April 1,

2006

• Time granularity of 15 minutes
• Official Kp database is essentially

uninterrupted since 1932

• Time granularity of 3 hours

Problem

Time granularity

Model predictions are made approximately every 15

minutes

Official Kp values are calculated once every 3 hours

Solution

Time-tag each of the official Kp values at the beginning of

the 3 hour interval and find model predictions that are made between 0 and 10 minutes after this time

Costello Validation

• Kp bins range from 0+ to 7+
• Figure 1: official Kp averages for each bin are plotted with error bars one standard

deviation in length

• Figure 2: the median official Kp values for each bin are plotted with error bars

showing the upper and lower quartiles

underprediction

• verprediction

underprediction

• verprediction

Solar Cycle Dependence

During solar

maximum external influences dominate activity in the magnetosphere

During solar

minimum internal dynamics are responsible for fluctuations in magnetic field strength

Solar Maximum Solar Minimum

Solar Cycle Dependence (Cont.)

Costello model appears to predict low Kp values slightly better during solar

maximum years

Solar Maximum Solar Minimum

Forecast Specific Validation

Figures show the distribution of official Kp values for Costello predictions

corresponding to NOAA warnings Expected Kp of 6 (G2 storm) Expected Kp of 7 or greater (G3 or higher storm)

Forecast Specific Validation (Cont.)

Figures show the distribution of official Kp values for Costello predictions

corresponding to NOAA warnings Expected Kp of 4 Expected Kp of 5 (G1 storm)

JHU/APL Models

APL Model 1

Inputs nowcast Kp and solar

wind parameters

Predicts Kp 1 hour ahead

APL Model 2

Same inputs as APL Model 1 Predicts Kp 4 hours ahead

APL Model 3

Inputs solar wind parameters Predicts Kp 1 hour ahead

APL Model 1

Inputs nowcast Kp and solar

wind parameters

Predicts Kp 1 hour ahead Correlation coefficient = 0.92

• verprediction

underprediction

APL Model 2

Inputs nowcast Kp and solar

wind parameters

Predicts Kp 4 hours ahead Correlation coefficient = 0.79

underprediction

• verprediction

APL Model 3

Inputs solar wind parameters Predicts Kp 1 hour ahead Correlation coefficient = 0.84

• verprediction

underprediction

Resolution to Discrepancy?

Interpolated Official Kp No Interpolation Interpolated Official Kp

Interpolation of official Kp

values may lead to skew in Wing’s validations

When no interpolation is

used, APL model tends to

• verpredict Kp instead of

underpredicting

Similar skew may be

responsible for discrepancy in Costello validations

No Interpolation

APL Model Validations

APL models installed Code edited to run on a

NOAA/SEC computer

Models successfully produce

real-time Kp estimates

Real-time data plots were not

produced

Modifications to run models

• ff of historical data were not

completed

Summary

• We found that the Costello model tends to overpredict Kp consistently
• Model performance may exhibit some solar cycle dependency
• Statistical evaluations will have to be performed in order to determine the extent of this

dependency

• Differences in performance are likely irrelevent for forecasting purposes
• Directly comparable validation studies should be carried out to determine if the

JHU/APL models perform significantly better than the Costello model

• Time interval, time granularity, and data set used should be identical

References

Detman, T., and J. A. Joselyn (1999), Real-time Kp predictions

from ACE real time solar wind, in Solar Wind Nine, edited by S. R. Habbal et al., AIP Conf. Proc., 471, 729-732.

Wing, S., J. R. Johnson, J. Jen, C.-I. Meng, D. G. Sibeck, K.

Bechtold, J. Freeman, K. Costello, M. Balikhin, and K. Takahashi (2005), Kp forecast models, J. Geophys. Res., 110, A04203, doi:10.1029/2004JA010500.

sd-www.jhuapl.edu/UPOS/ www.gfz-potsdam.de www.n3kl.org www.ngdc.noaa.gov www.sec.noaa.gov

Acknowledgements and Thanks

NOAA/SEC

Douglas Biesecker Christopher Balch

JHU/APL

Simon Wing Janice Schofield